|Publication number||US6941007 B1|
|Application number||US 09/493,663|
|Publication date||Sep 6, 2005|
|Filing date||Jan 28, 2000|
|Priority date||Jan 28, 2000|
|Also published as||US20060002606|
|Publication number||09493663, 493663, US 6941007 B1, US 6941007B1, US-B1-6941007, US6941007 B1, US6941007B1|
|Inventors||Douglas D. Do|
|Original Assignee||Micron Technology, Inc.|
|Export Citation||BiBTeX, EndNote, RefMan|
|Patent Citations (11), Referenced by (8), Classifications (12), Legal Events (5)|
|External Links: USPTO, USPTO Assignment, Espacenet|
1. Field of the Invention
The present invention relates to the field of semiconductor fabrication and, more particularly to a method and apparatus that uses multiple images in a pattern recognition process used to detect defects in the manufacture of a semiconductor device.
2. Description of the Related Art
In the semiconductor industry, there is a continuing movement towards higher integration, density and production yield, all without sacrificing throughput or processing speed. The making of today's integrated circuits (ICs) requires a complex series of fabrication, inspection and testing steps interweaved throughout the entire process to ensure the proper balance between throughput, processing speed and yield. The inspections and tests are designed to detect unwanted variations in the wafers produced, as well as in the equipment and masks used in the fabrication processes. One small defect in either the devices produced or the process itself can render a finished device inoperable.
Many of the inspection steps once done manually by skilled operators have been automated. Automated systems increase the process efficiency and reliability as the machines performing the inspection are more consistent than human operators who vary in ability and experience and are subject to fatigue when performing repetitive tasks. The automated systems also provide greater amounts of data regarding the production and equipment, which enables process engineers to both better analyze and control the process.
One such automated inspection step is known as pattern recognition or pattern inspection. Many different “patterns” appear on both the wafer and the masks used to produce the ICs. Typical pattern inspection systems are image based, as described, for example, in U.S. Pat. Nos. 4,794,646; 5,057,689; 5,641,960; and 5,659,172. In U.S. Pat. No. 4,794,646, for example, the wafer, or part thereof, is scanned and a highly resolved picture or image of the pertinent “pattern” is obtained. This pattern image is compared to other pattern images retrieved from the same or other wafers, or is compared to an ideal image stored in the inspection system database. Differences highlighted in this comparison identify possible defects in the IC or wafer.
The method 10, however, is not without its shortcomings. Referring to
As can be seen from
As can be seen from
Typically, the inspection apparatus will allow a user to set pattern recognition thresholds. These thresholds are designed to reduce or increase the matching percentage required between the pattern to be recognized and the inspected images. Thus, a user may set a matching threshold to 100%, in which case, the apparatus will only declare matches when the inspected images contain patterns that exactly match the pattern to be recognized. This would ensure that defective images 54 a, 54 b, 54 c are not matched to desired and non-defective images. However, due to variations in the manufacturing process, a matching threshold of 100% would most likely lead to no desired matches or too few desired matches than are actually present. The apparatus would not detect all of the proper matches, if it detects any at all (i.e., it is under inclusive). On the other hand, if the matching threshold is set too low, e.g., 50%, then too many matches will occur. These matches will include defective images 54 a, 54 b, 54 c whose patterns are within the matching percentage (i.e., it is over inclusive). Typically, the matching threshold is set to approximately 65% to balance between the over inclusive and under inclusive matching thresholds.
Even with a threshold setting of 65%, the conventional pattern recognition process is still unreliable.
It would be desirable to use the fourth sample image 96 as the pattern to be detected. As noted above, the fourth image 96 does not contain any top surfaces 52. However, the fourth sample image 96, which has bottom surfaces 54 substantially matching the bottom surfaces 54 of the first sample image 90, would not match any of the other images 90, 92, 94 because the other images contain both top and bottom surfaces 52, 54. Thus, even if it were possible to select the fourth sample image 96 as a pattern to be recognized, the pattern recognition analysis would be corrupted by the top surfaces 52, of the inspected images (e.g., images 90, 92, 94) which are not part of the desired features.
Accordingly, there is a desire and need for a pattern recognition process that filters out undesirable features from the object being inspected prior to performing a pattern recognition analysis on the object. There is also a desire and need for a pattern recognition process that allows a user to select multiple desired images of the object being inspected to be used as a pattern to be recognized during a pattern recognition analysis on the object.
The present invention provides a pattern recognition technique that substantially filters out undesirable features of the object being inspected prior to performing a pattern recognition analysis on the object.
The present invention also provides a pattern recognition technique that allows multiple desired images of the object being inspected to be used as a pattern to be recognized during a pattern recognition analysis on the object.
The above and other features and advantages of the invention are achieved by a pattern inspection apparatus and method that uses multiple images in a pattern recognition process used to detect defects in an object being inspected. A user is provided with multiple image selection windows allowing the user to select multiple desired images from the object to form a pattern to be recognized within the object. The multiple desired images will be substantially free from undesired features of the object. Once the multiple desired images are selected, the relationship between them is determined and used to learn the pattern to be recognized. The relationships between the desired images further filters out undesired features. The pattern to be recognized is used in a subsequent pattern recognition analysis. Since the pattern to be recognized includes only desired images and their relationship, undesired features that could corrupt the pattern recognition analysis are not present during the analysis. Thus, the apparatus and method are more accurate than prior inspection tools.
The foregoing and other advantages and features of the invention will become more apparent from the detailed description of the preferred embodiments of the invention given below with reference to the accompanying drawings in which:
In an exemplary implementation, a pattern inspection or defect detection apparatus 100, as shown in
The SEM 102 is provided with an object support structure 108 in the form of a stage or chuck, which may be moveable or stationary. An object 120 under evaluation such as an IC sample die or a wafer containing many IC dies rests on support structure 108. Under control of computer 110, the structure 108 may be moveable in two (X-Y) or three (X-Y-Z) dimensions to facilitate the proper viewing of the object 120 (or parts thereof). A deflector 104 and detector 106, whose operations will be described in detail below, are also provided within the SEM 102 to assist in the viewing of the object 120. An image processor 112, together with its accompanying image memory 114 are provided to process the image signals output by the SEM 102 and transform the signals into visual representations or data which can be viewed on a display monitor 116 (e.g., cathode ray tube (CRT)) or used for processing in computer 110.
In operation, the SEM 102 uses a finely focused electron beam directed by deflector 104, preferably under the control of the computer 110, to scan the surface of the object 120 resting on the support 108, typically in two dimensions (X-Y). For the purposes of discussion only, it will be assumed herein that the object 120 under evaluation is a silicon wafer having a plurality of contacts formed therein, such as contacts 50 illustrated in
The detected electron current is output as an image signal to the computer 110 and image processor 112 where an image representative of the surface of the object 120 can be formed based on the image signal. This image may be stored in the image memory 114 and can be viewed on the monitor 116 or otherwise processed by the computer 110. The high resolution of the image is attributed to the small diameter (e.g., several nanometers) of the electron beam illuminator. The visual contrast achieved in the image originates mostly from variations in the extent of the secondary electron emissions from the topographic features of the surface of the object 120.
The apparatus 100 then prompts the user to input the number of image selection windows required (step 206). Each image selection window is used by the method 200 to select an image, within the inspected images, to be used as part of the pattern to be detected. In the prior art method one large box is used to obtain an image having both desired (e.g., bottom surfaces of contacts) and undesired features (e.g., top surfaces of contacts). The present method 200 uses multiple sizable image selection windows. This way, instead of placing one large box over an image containing both desired and undesired features, the method 200 uses several multiple sizable image selection windows to select images with only desired features. As will become apparent below, this allows the method 200 to substantially filter out undesired features from the pattern to be recognized and thus, the pattern recognition analysis.
At step 208, the user inputs the number of desired windows. By allowing the user to select more than one image selection window, the method 200 of the present invention allows multiple different images to be used in the pattern recognition analysis. It should be noted that these steps differ from the prior art, which only provides the user with one box, and thus, does not allow the user to select multiple images (see step 16 of
Referring also to
At step 214, the apparatus 100 learns the selected images and the relationship between them. That is, the apparatus 100 obtains and stores information for the selected images. The information can be stored in the memory 114 or other computer readable medium connected to or contained within the apparatus 100. The spatial relationship between the selected images is also obtained and stored. Learning the spatial relationship between the images will further filter out undesired features. This can best be illustrated by the following example. Referring to
Referring again to
If at step 412, it is determined that more image selection windows are required, the method 400 continues at step 406 where a new image selection window is displayed on the field of view. This way, the user may select multiple images using multiple image selection windows. By using multiple image selection windows, the user can size these windows so that they only select desired features. That is, it is possible for the user to substantially filter out unwanted features, such as the top surfaces of the contacts, from the pattern to be recognized.
If at step 412 it is determined that no more image selection windows are required, the method continues at step 414. At step 414, the apparatus 100 learns the selected images and the relationship between them. That is, the apparatus 100 obtains and stores information for the selected images. The information can be stored in the memory 114 or other computer readable medium connected to or contained within the apparatus 100. The spatial relationship between the selected images is also obtained and stored (as described above with reference to
The present invention improves the pattern recognition process by allowing the user to create a pattern to be recognized using multiple images. By select multiple images, the user can substantially filter out unwanted features, such as the top surfaces of contacts formed in a wafer. More importantly, the present invention uses relationships between the selected images to hone in on the exact pattern to be recognized. Thus, the present invention can perform pattern recognition using the typical thresholds currently used in today's inspection systems, but with substantially better results.
It should be appreciated that the learned pattern created by the methods of the present invention can be used in the pattern recognition analysis to detect defects in the object as well as to detect desired (i.e., properly etched) patterns in the object. It should also be appreciated that the learned pattern created by the methods of the present invention can be used in both pattern recognition and critical dimension (CD) analysis processes. Moreover, the methods of the present invention can be used to inspect wafers, reticles or other semiconductor devices requiring pattern inspection/recognition.
While the invention has been described in detail in connection with the preferred embodiments known at the time, it should be readily understood that the invention is not limited to such disclosed embodiments. Rather, the invention can be modified to incorporate any number of variations, alterations, substitutions or equivalent arrangements not heretofore described, but which are commensurate with the spirit and scope of the invention. Accordingly, the invention is not to be seen as limited by the foregoing description, but is only limited by the scope of the appended claims.
|Cited Patent||Filing date||Publication date||Applicant||Title|
|US4579455 *||May 9, 1983||Apr 1, 1986||Kla Instruments Corporation||Photomask inspection apparatus and method with improved defect detection|
|US4794646||Aug 14, 1986||Dec 27, 1988||Mitsubishi Denki Kabushiki Kaisha||Charged beam pattern defect inspection apparatus|
|US5057689||Sep 17, 1990||Oct 15, 1991||Matsushita Electric Industrial Co., Ltd.||Scanning electron microscope and a method of displaying cross sectional profiles using the same|
|US5568563||May 3, 1994||Oct 22, 1996||Mitsubishi Denki Kabushiki Kaisha||Method and apparatus of pattern recognition|
|US5641960||Nov 4, 1994||Jun 24, 1997||Fujitsu Limited||Circuit pattern inspecting device and method and circuit pattern arrangement suitable for the method|
|US5659172||Feb 23, 1996||Aug 19, 1997||Opal Technologies Ltd.||Reliable defect detection using multiple perspective scanning electron microscope images|
|US5764793||Dec 29, 1995||Jun 9, 1998||Dainippon Screen Mfg. Co., Ltd.||Method of and apparatus for inspecting pattern defects|
|US5822055||Jun 27, 1997||Oct 13, 1998||Kla Instruments Corporation||Optical inspection of a specimen using multi-channel responses from the specimen using bright and darkfield detection|
|US5917588 *||Nov 4, 1996||Jun 29, 1999||Kla-Tencor Corporation||Automated specimen inspection system for and method of distinguishing features or anomalies under either bright field or dark field illumination|
|US6178223 *||Oct 6, 1998||Jan 23, 2001||Cardiac Mariners, Inc.||Image reconstruction method and apparatus|
|US6483937 *||Jun 17, 1999||Nov 19, 2002||International Business Machines Corporation||Process for inspecting an object|
|Citing Patent||Filing date||Publication date||Applicant||Title|
|US7702158 *||Oct 21, 2003||Apr 20, 2010||Keyence Corporation||Image processing apparatus, image processing method and record medium for the image processing apparatus|
|US7783113 *||Oct 8, 2004||Aug 24, 2010||Drvision Technologies Llc||Partition pattern match and integration method for alignment|
|US7804993||Feb 28, 2005||Sep 28, 2010||Applied Materials South East Asia Pte. Ltd.||Method and apparatus for detecting defects in wafers including alignment of the wafer images so as to induce the same smear in all images|
|US7813541 *||Feb 28, 2005||Oct 12, 2010||Applied Materials South East Asia Pte. Ltd.||Method and apparatus for detecting defects in wafers|
|US8031931||Apr 24, 2006||Oct 4, 2011||Applied Materials South East Asia Pte. Ltd.||Printed fourier filtering in optical inspection tools|
|US20040136588 *||Oct 21, 2003||Jul 15, 2004||Kazuhito Saeki||Image processing apparatus, image processing method and record medium for the image processing apparatus|
|US20060078192 *||Oct 8, 2004||Apr 13, 2006||Seho Oh||Partition pattern match and integration method for alignment|
|US20060193506 *||Feb 28, 2005||Aug 31, 2006||Negevtech Ltd.||Method and apparatus for detecting defects in wafers including alignment of the wafer images so as to induce the same smear in all images|
|U.S. Classification||382/145, 382/144, 382/143|
|International Classification||G06K9/62, G06K9/00, G06T7/00|
|Cooperative Classification||G06K9/6255, G06T7/001, G06K9/6253|
|European Classification||G06T7/00B1R, G06K9/62B6, G06K9/62B5|
|Jan 28, 2000||AS||Assignment|
Owner name: MICRON TECHNOLOGY, INC., IDAHO
Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNOR:DO, DOUGLAS D.;REEL/FRAME:010538/0083
Effective date: 20000121
|Feb 4, 2009||FPAY||Fee payment|
Year of fee payment: 4
|Feb 6, 2013||FPAY||Fee payment|
Year of fee payment: 8
|May 12, 2016||AS||Assignment|
Owner name: U.S. BANK NATIONAL ASSOCIATION, AS COLLATERAL AGEN
Free format text: SECURITY INTEREST;ASSIGNOR:MICRON TECHNOLOGY, INC.;REEL/FRAME:038669/0001
Effective date: 20160426
|Jun 2, 2016||AS||Assignment|
Owner name: MORGAN STANLEY SENIOR FUNDING, INC., AS COLLATERAL
Free format text: PATENT SECURITY AGREEMENT;ASSIGNOR:MICRON TECHNOLOGY, INC.;REEL/FRAME:038954/0001
Effective date: 20160426